[clang] [CIR] Vector saturating rounding shift right (SISD variants) (PR #200113)

[Andrzej Warzyński via cfe-commits]

================
@@ -29,6 +29,66 @@
 // TODO: Implement SISD variants
 //===------------------------------------------------------===//
 
+// ALL-LABEL: @test_vqrshrunh_n_s16(
+uint8_t test_vqrshrunh_n_s16(int16_t a) {
+  // CIR:  {{%.*}} = cir.call_llvm_intrinsic "aarch64.neon.sqrshrun"
+
+  // LLVM-SAME: i16 noundef {{.*}})
----------------
banach-space wrote:

Here and in other places:
1. Don't check the attributes.
2. Track the input variables (i.e. the call to the intrinsics should be using the input argument). 
```suggestion
  // LLVM-SAME: i16 {{.*}}) [[A:%.*]])
```

https://github.com/llvm/llvm-project/pull/200113